From July to October 2022, samples (Silage corn, Silage alfalfa, and animal feces) were collected from the Qinghai-Tibet Plateau region, including Diebu County, Xiahe County, Gaotai County and Sunan County, Tianzhu Tibetan Autonomous County in Gansu Province. Xining City, Haixi Mongolian and Tibetan Autonomous Prefecture in Qinghai Province. Each 25 g sample was thoroughly mixed with 225 mL of sterile physiological saline solution. Subsequently, a gradient dilution was performed to achieve dilution levels of 10⁻³~10⁻⁷. All dilutions were evenly spread onto a culture medium with coumarin as the sole carbon source (CM medium; L-1distilled water): 10 g coumarin, 0.25 g KH₂PO₄, 1 g NH₄NO₃, 1 g CaCl₂, 0.25 g MgSO₄·7H₂O, 1 mg FeSO₄, and 15 g agar (Rao et al., 2017). The cultures were then incubated at 37°C for 72 h. Colonies displaying growth on the CM plates were chosen for the degradation and rescreening experiments of AFB₁.

The preparation method for TSB-AFB₁ liquid medium was as follows: Take 1 mL of AFB₁ standard solutions (Sigma-Aldrich, St. Louis, MO, United States) with concentrations of 200, 400, 600, 800, 1,000 μg/mL, and add them to 100 mL of TSB liquid medium, respectively. The final AFB₁ concentrations in TSB-AFB₁ liquid medium were 2, 4, 6, 8, 10 μg/mL, with the pH adjusted to 6.8. Following a modified method based on Chen et al. (2022), a 10 mL bacterial suspension with a concentration of 10⁸ CFU/mL was inoculated into 100 mL TSB-AFB₁ liquid medium (with AFB₁ concentrations of 2, 4, 6, 8, 10 μg/mL), and the cultures were shaken culture (37°C, 180 rpm, 72 h). A sterile TSB-AFB₁ was used as a control. The second screening of initial strains was based on their ability to degrade AFB₁ in TSB-AFB₁ medium with a concentration of 10 μg/mL AFB₁. Evaluated the degradation ability of the finally selected strains toward varying concentrations of AFB₁ using TSB-AFB₁ medium with different concentrations (2, 4, 6, 8, 10 μg/mL).

To 5 mL of the test solution, 20.0 mL of acetonitrile-water solution (V:V/84:16) was added. After shaking for 20 min and centrifugation (8,000 rpm, 8 min), 4 mL of the supernatant was taken and subjected to three consecutive extractions with an equal volume of chloroform. The extract was evaporated under nitrogen at 55°C. The precipitate was dissolved in 1 mL of DMSO, filtered through a 0.22 μm organic membrane, and the filtrate was analyzed for AFB₁ content using High-Performance Liquid Chromatography (HPLC) system (Waters Acquity, Milford, MA, United States).

HPLC conditions for AFB₁ detection: Equipped with a BEH C18 chromatographic column (100 mm × 2.1 mm, 1.7 μm) and a 360 nm ultraviolet detector. Column temperature: 40°C; mobile phase: acetic acid ammonium/methanol; injection volume: 10 μL; flow rate: 0.2 mL/min.

In both cases, the percentage of AFB₁ degradation was calculated by following formula, where C was the AFB₁ Paek area in treatment, and F was the AFB₁ peak area in control:

The maximum value of AFB₁ contamination in food was 10 μg/g, we artificially contaminated real food samples to reach this concentration. Experimental samples applied easily contaminated foods like maize, cheese, and peanuts. To prepare these artificially contaminated samples, 10 mL of AFB₁ standard solution (1,000 μg/mL) was added separately to 1 kg of maize, cheese, and peanuts. Thorough mixing ensured a final AFB₁ concentration of 10 μg/g. Following this, each 1 kg artificially contaminated food sample received a uniform spray of 100 mL bacterial suspension (10⁸ CFU/mL), followed by shading incubation (37°C, 72 h). Treatment groups comprised M-YUAD7, C-YUAD7, and P-YUAD7. Positive controls included food samples artificially contaminated with AFB₁ but without the inoculated strain (C-M, C-C, C-P). Negative controls consisted of naturally incubated food samples (N-M, N-C, N-P), and strains were inoculated into food samples without artificial AFB₁ contamination (UnM-YUAD7, UnC-YUAD7, UnP-YUAD7). All control groups were also incubated without light (37°C, 72 h). The pretreatment steps for detecting AFB₁ content in actual food samples were as follows: 20 g of the sample was mixed with 180 mL distilled water in a juice extractor, blended at high speed for 30 s, and subsequently filtered through four layers of gauze. The obtained pulverized extract will be retained for subsequent AFB₁ content detection, following the same procedures as in step 2.2.

For physiological and biochemical identification of the target strain, a conventional microbial biochemical identification kit (Beijing Luqiao, Beijing, China) was employed, following the guidelines of “Bergey’s Manual of Systematic Bacteriology” (Garrity, 1994).

The genomic DNA of YUAD7 was extracted via a SanPrep DNA purification kit (Sangon Biotech, Shanghai, China), following the manufacturer’s guidelines. Subsequently, a combination of PacBio RS II Single Molecule Real Time (SMRT, Pacific Biosciences, MenloPark, CA, United States) and Illumina sequencing platforms (Illumina Novaseq 6000, Shanghai, China) were employed for sequencing. The accession number PRJNA964696 at the US National Center for Biotechnology Information (NCBI) was assigned to the sequence data of B. amyloliquefaciens YUAD7.

The CDSs were predicted with gene annotation performed using GO and KEGG (Fang et al., 2020) using sequence alignment tools such as BLAST,¹ Diamond (Version 0.8.35) and HMMER.² All data were quantified and visualized on the CGView data platform³ and the Chiplot online platform.⁴

The AFB₁ degradation capabilities of cell-free supernatant, intracellular extracts, and dead-cell bacterial suspensions were evaluated using a previously described method (Rao et al., 2017; Cai et al., 2020). The cell-free supernatant, intracellular extracts, and bacterial suspension of dead cells were incubated with TSB-AFB₁ medium (AFB₁ concentration of 10 μg/mL) at 37°C with 180 rpm shaking for 72 h. Control cultures consisted of TSB medium or sterile phosphate buffer supplemented (PBS) with AFB₁(concentration of 10 μg/mL). The residual AFB₁ was then analyzed as previously described.

The degradation of AFB₁ active components in target strains was fractionated into four components, and the effects of proteinase K (PK, 1 mg/mL), sodium dodecyl sulfate (SDS, 1%), PK + SDS, and heat treatment (boiled for 20 min) on the degradation of AFB₁ were investigated. Subsequently, each component was incubated with sterile PBS containing AFB₁ at a 10 μg/mL concentration at 37°C. A sterile PBS with AFB₁ concentration at 10 μg/mL was used as a control. After 72 h, residual AFB₁ was detected, as described earlier.

The survival monitoring assay utilized L-02 standard cells obtained from the Cell Preservation Bank at the Chinese Academy of Sciences in Shanghai, representing a human normal liver cell line. The NC group was cultured in RPMI 1640 medium, while the CC group was cultivated in AFB₁-RPMI 1640 solution (AFB₁ concentration at 10 μg/mL). The EG group was exposed to RPMI 1640 medium with 10% degradation solution. Then, L-02 cells were seeded at a density of 100,000 cells per well in a 24-well plate and incubated in RPMI 1640 medium at 37°C for 24 h under a 5% CO₂ atmosphere to synchronize the population. Subsequently, the medium was replaced with fresh medium containing the test mentioned above samples. The survival monitoring assay was performed using the Vi-cell system (Thermo Scientific, Waltham, MA), and the cell morphology of each sample was compared with that of its corresponding control.

To assess the mutagenic potential of the degradation products, the Ames test using the Genotoxic Ames kit (Iphase Bio Technology, Suzhou, China) following the manufacturer’s instructions and the protocol outlined by Chen et al. (2022). The degradation products obtained from a 72-h co-culture with the AFB₁-degrading bacterium and AFB₁ were incubated with Salmonella Typhimurium TA100 or TA102 at 37°C for 48 h (EG). The count of S. typhimurium colonies was documented, and the results were expressed as the number of revertant colony-forming units (CFUs). Positive controls consisted of samples extracted from TSB medium with AFB₁(CC), while negative controls included extracts from TSB medium without AFB₁(NC).

The degradation liquid was concentrated to afford a crude residue which was suspended in H₂O and then extracted with petroleum ether, EtOAc, and n-BuOH to afford fractions.

The petroleum ether fraction (50 g) subjected to silica gel column chromatography (200–300 mesh, Qingdao Haiyang Chemical Factory, Qingdao, China) with DCM and MeOH (1:0~0:1, v/v) gradient elution was distributed as two fractions (Fr1, 15 g, Fr2 6.3 g). Fr1(15 g) was chromatographed on an ODS with MeOH/H₂O (0:1~0:1, v/v), and then separated by preparative HPLC (MeOH/H₂O, 40:60, V/V; flow rate, 4 mL/min) to yield compounds 1 (tR = 8.7 min, 20.4 mg), 2 (tR = 6.8 min, 14.6 mg).

Fr2(6.3 g) was separated by silica gel column chromatography (cc, 70 × 245 mm) with DCM/MeOH (1:0 ~ 0:1, V/V) to provide two subfractions (Fr2-1, 1.3 g; Fr2-2, 0.8 g). Fr2-1(1.3 g) was further separated by preparative HPLC (MeOH/H₂O, 35:65, V/V; flow rate, 4 mL/min) to yield 3 (tR = 7.6 min, 26.3 mg), yield 4 (tR = 5.4 min, 10.5 mg).

The UPLC-Q-Orbitrap included a UPLC system equipped with an autoinjector and quaternary UPLC pump (Waters Acquity, Milford, MA, United States), and quadrupole/electrostatic field orbitrap high-resolution mass spectrometry (Q-Orbitrap HRMS; Thermo Scientific, Waltham, MA). The chromatograph was equipped with a BEH C18 column (100 mm × 2.1 mm, 1.7 μm) and operated at a column temperature of 40°C. Samples were injected and eluted using a mobile solvent containing mobile phase A, which was an aqueous solution containing 0.1% formic acid and 5 mmol/L ammonium formate, and mobile phase B, a methanol solution containing 0.1% formic acid and 5 mmol/L ammonium formate. The mass spectrometry parameters were set similar to that in previous study (Lai et al., 2023).

The ¹H and ¹³C NMR experiments data were obtained by Bruker DPX-400 spectrometer in dimethyl sulfoxide-d6 (DMSO-d6; DPX-400, Bruker, Switzerland). The sample was equipped with a 5 mm NORELL NMR tube and operated at 25°C. The data was analyzed using MestreNova software (ver. 14.2, Mestrelab Research, Escondido, CA). Then, BioTransformer 3.0 software⁵ was used to predict the AFB₁ degradation pathway of YUAD7 based on the structure of degradation products.

The results were presented as mean ± standard deviation (SD). Statistical analysis was performed using SPSS software and involved ANOVA followed by Duncan’s test. Different lowercase letters in the bars of each group indicated significant differences between treatments (p < 0.05).

As shown in Table 1, 23 strains capable of utilizing coumarin were obtained using coumarin as the sole carbon source for enrichment and acclimation. YUAD7 exhibited the highest degradation rate, achieving a 91.7% degradation within 72 h in TSB-AFB₁ solution with an AFB₁ concentration of 10 μg/mL. Therefore, YUAD7 was chosen as the target strain for further research.

YUAD7’s ability to degrade AFB₁ at different concentrations was assessed under consistent incubation time and temperature condition. As depicted in Figure 1A, When the AFB₁ concentration was below 6 μg/mL, apart from the initial 24-h, the degradation percentages of AFB₁ by YUAD7 were consistently similar across treatments during the 96-h incubation. The degradation rates nearly reached their maximum at 72-h, exceeding 99%. Furthermore, when the AFB₁ concentrations were 8 and 10 μg/mL, YUAD7 exhibited time-dependent degradation, achieving a reduction of over 94% by 96-h (AFB₁ 10 μg/mL). Considering the maximum level in raw cereal grains (Campos-Avelar et al., 2021), 10 μg/mL AFB₁ was chosen for the subsequent research.

YUAD7 was employed to degrade AFB₁ in real food samples, as shown in Figure 1B. AFB₁ levels in all negative control treatments were below the limit of detection [LOD: 0.003 μg/g (mL)]. In the positive controls of maize, cheese, and peanuts, AFB₁ levels were 10.84 ± 0.17, 10.73 ± 0.35, and 9.83 ± 0.26 μg/g, respectively. After inoculating YUAD7 for AFB₁ degradation, the levels in maize, cheese, and peanuts were reduced to 1.62 ± 0.08, 1.43 ± 0.06, and 1.24 ± 0.05 μg/g, respectively. When AFB₁ contamination in food reached 10 μg/g, YUAD7 effectively removed over 85% of AFB₁ from real food samples within 72 h.

Strain YUAD7 morphological, characteristics physiological and biochemical characteristics were shown in Supplementary Figure 1 and Supplementary Table 1. Referencing Bergey’s Manual of Systematic Bacteriology (Garrity, 1994), the strain YUAD7 was preliminarily identified as belonging to the Bacillus.

Whole genome sequencing was performed on strain YUAD7, and the results were shown in Supplementary Table 2 and Figure 2A. The complete genome of YUAD7 was a circular chromosome with a size of 4,028,188 bp and a G + C content of 46.4%. The DNA coding sequences encompassed 3,628,166 bp, constituting approximately 90.07% of the total genome. Following genome assembly and prediction annotation, a total of 4,172 predicted genes were identified, of which 4,054 were CDSs and 118 were RNA genes. Upon manual prediction and annotation, 3,218 and 2,106 predicted genes were annotated in the GO and KEGG databases. Based on the analysis of 16S rRNA and 30 housekeeping genes, including acnA, dacB, licR, mfd, lplJ, lpdA, lepA, secA, budA, atpA, gltB, bamA, rfbF, rpoC, rpoB, leuS, dnaE, thrS, pheT, hsdR, recJ, pbpA, SerA, dinG, mobl, rarD, gltB, xynB, pyc, and lpdA, the closest relative of YUAD7 was Bacillus. amyloliquefaciens strain (Supplementary Figure 2). Consequently, the strain mentioned above was designated as Bacillus. amyloliquefaciens YUAD7.

YUAD7’s CDSs annotation information in the GO and KEGG databases was shown in Figures 2B,C, respectively. Figure 2B indicated that the annotation information of YUAD7’s encoded genes in the GO database mainly fell into three categories: biological process, cellular component, and molecular function, such as oxidation-dependent protein catabolic process (GO: 0070407), polysaccharide biosynthetic process (GO: 0000271), carbohydrate catabolic process (GO: 0044193), hydroxyisourate hydrolase complex (GO: 0106232), O6-methyl-dGTP hydrolase activity (GO: 0106433), and stearyl deacetylase activity (GO: 0034084). Furthermore, in Figure 2C, KEGG annotation information demonstrated that the functions annotated by gene sequences in the primary metabolic pathways could be divided into four categories: cellular processes, environmental information processing, genetic information processing, and metabolism. These findings provide an understanding of the functional genes and metabolic pathways involved in AFB₁ degradation by B. amyloliquefaciens YUAD7 at the genomic level. These predicted metabolic pathways and functional genes were closely associated with the process of AFB₁ degradation by the strain.

All the data indicated that B. amyloliquefaciens YUAD7 could serve as a beneficial and safe bacterium to be applied in food and feed processing.

In order to investigate the mechanism of AFB₁ removal by B. amyloliquefaciens YUAD7, the study tested the efficiency of cell-free supernatant, intracellular extracts, and dead cells in degrading AFB₁. As shown in Figure 3A, after 72-h of cultivation, cell-free supernatant could remove 75.42% ± 3.09% of AFB₁ (10 μg/mL), while the removal rates for dead cells and intracellular extracts were 4.36% ± 0.61% and 12.63% ± 2.05%, respectively. The cell-free supernatant of B. amyloliquefaciens YUAD7 was more effective in reducing AFB₁ than dead cells and intracellular extracts (p < 0.05). These results suggested that the removal of AFB₁ by B. amyloliquefaciens YUAD7 was mainly dependent on degradation, and the cell-free supernatant was the main active component in the AFB₁ degradation process. SDS, proteinase K (PK), and PK + SDS effected on the activity of YUAD7’s cell-free supernatant in degrading AFB₁ was shown in Figure 3B, AFB₁ degradation capacity of the cell-free supernatant rated down to 52.85% ± 2.25%, 13.26% ± 2.88%, and 7.36% ± 0.95%. The result indicated that the AFB₁ degradation agents in B. amyloliquefaciens YUAD7 cell-free supernatant included not only enzymes or proteins but also other non-protein components. Additionally, the cell-free supernatant still exhibited AFB₁ degradation activity of 43.16 ± 3.54% even after boiling for 20 min (Figure 3B).

The study examined the impact of AFB₁ and its degradation products by B. amyloliquefaciens YUAD7 on the lifespan of L-02 cells. As shown in Figure 4A, there was no statistical difference (p > 0.05) in cell viability between the EG group and the NC group, with L-02 cell survival rates exceeding 92%. However, the CC group exhibited a 62.5% decrease in the average lifespan of L-02 cells. The morphological changes of L-02 cells under different treatment conditions within 72-h of cultivation were shown in Figure 4B. Compared to the NC group, the EG group showed no significant differences in cell morphology (p > 0.05). In contrast, the CC group exhibited cell elongation, lysis phenomena, and a significant reduction in cell viability. These results showed that B. amyloliquefaciens YUAD7 degraded AFB₁ into metabolites without toxicity to the L-02 cells. Meanwhile, the Ames test was employed to evaluate the mutagenic potential of AFB₁ degradation products facilitated by B. amyloliquefaciens YUAD7. Compared to the control group, a roughly twofold increase in the count of revertant CFUs from S. typhimurium TA100 and TA102 was noted in the AFB₁ group (CC). However, there was no statistically significant difference in revertant CFUs between the degradation products (EG) and the control group (NC; Figure 4C), indicating that B. amyloliquefaciens YUAD7 transformed AFB₁ into metabolites with diminished mutagenicity.

Chemical components of AFB₁ degradation products by strain YUAD7 were analyzed using the UPLC-Q-Orbitrap HRMS method. The characteristic ions of AFB₁ standard compounds, which were m/z 213, 128, 115, 77, 69, and 43, respectively (Figure 5A). Comparing the fragment ions of AFB₁ and the compounds in degradation solution, compound 1–4 had high homology with the fragment ions of AFB₁, and compound 1–4 were the products of the degradation of AFB₁ by the B. amyloliquefaciens YUAD7 (Supplementary Figure 3).

Compound 1 was isolated as white powder. Its molecular formula was established as C₁₂H₁₄O₄ by UPLC-Q-Qrbitrap HRMS (m/z 222.0892), sharing several homologous fragment ions with AFB₁, such as 213, 128, 115, 77, 69, 43, etc. (Figure 5B). Its ¹H NMR spectrum (Table 2) showed characteristic signals for the hydrogen of ABX coupling system on the benzene ring [δ_H7.34 (1H, d, J = 8.2 Hz, H-6), 6.75(1H, dd, J = 8.1, 2.3 Hz, H-1), and 6.52(1H, d, J = 2.2 Hz, H-3)]; δ_H3.80(3H, s, H-7), 2.42(3H, s, H-9),3.86(3H, s, H-12)] were three methoxy hydrogen signals. δ_H6.14 (1H, q, J = 1.3 Hz, H-10) was the hypomethyl hydrogen signal. Analysis of the ¹³C NMR data (Table 2) revealed 12 carbon signals, including 6 carbon signals on the benzene ring, three methoxy carbon signals (δ_C55.67, 20.44, 55.92), and three sp³ hybridized carbon data[δ_C153.39(C-8), 115.33(C-10), 170.01(C-11)] which indicated that compound 1 had the basic skeleton of dimethoxy benzene. The determination of the -CH linkage position and its sequential arrangement within compound 1 was determined by HSQC spectra (Supplementary Figure 6). The benzene ring structure suggested by HSQC correlations from δ_H7.34 (1H, d, J = 8.2 Hz, H-6) to δ_C129.36(C-6), δ_H6.75(1H, dd, J = 8.1, 2.3 Hz, H-1) to δ_C107.87(C-1), δ_H6.52(1H, d, J = 2.2 Hz, H-3) to δ_C97.85(C-3). The three methyl groups were at C7, C9 and C12 positions, respectively, [δ_H3.80 (3H, s) to δ_C55.67(C-7), δ_H2.42 (3H, s) to δ_C20.44(C-9), δ_H3.86 (3H, s) to δ_C55.92(C-12). Finally, the hydrocarbon formation pertaining to compound 1 was precisely ascribed through the chemical formula, ¹H NMR, ¹³C NMR, HSQC, and the structure of compound 1 was (2-4-dimethoxyphenyl) but-2-enoic acid (Figure 6A).

Compound 2 was isolated as a yellow powder with the molecular formula C₅H₁₂N₂O₂, which was established by UPLC-Q-Orbitrap HRMS (m/z 132.0898), sharing several homologous fragment ions with AFB₁ such as 128, 115, 77, 69, 43, etc. (Figure 5C). The ¹H NMR data (Table 2) of 2 showed the hydrogen signal of two methoxies [δ_H 2.38 (3H, s, H-1″, H-1a´´)]. δ_H 2.89 (2H, t, J = 6.7 Hz, H-2″) and 4.26 (2H, t, J = 6.7 Hz, H-3″) were two hydrogen signals on methylene, and δ_H 4.88 (2H, s, H-4″) was the nitrile hydrogen signal. ¹³C NMR spectra (Table 2) showed the presence of 5 carbon signals. Among them, 2 signals belonged to 2 methoxies [δ_C 45.89 (C-1″) and (C-1a″)]. δ_C 58.52 (C-2″) and 62.73(C-3″) were the methylene carbon signal. δ_C157.14 (C-4″) were the aminoester group carbon signal. Based on the above date, as well as HSQC data (Supplementary Figure 9), 2 was speculated 2-(dimethylamino) ethyl carbamate (Figure 6B).

Compound 3 was obtained as a yellow powder and the UPLC-Q-Orbitrap HRMS data of 3 (m/z 166.0993, calcd for C₁₀H₁₄O₂) indicated its molecular formula of C₁₀H₁₄O₂, sharing several homologous fragment ions with AFB₁ such as 128, 115, 77, 69, 43, etc. (Figure 5D). In the ¹H NMR data (Table 2) δ_H 1.67–1.61(3H, m, H-1′, H-3′) were the characteristic hydrogen signals of the methoxy. δ_H 5.38 (1H, th, J = 6.6, 1.6 Hz, H-4′) and 6.51(1H, tq, J = 5.9,1.4 Hz, H-6′) were two methenyl hydrogen signals. δ_H 2.74 (2H, dddq, J = 7.1, 6.2, 2.0, 1.0 Hz, H-5′) and 3.00 (2H, dp, J = 5.8,1.0 Hz, H-8′) were two hydrogen signals on methylene. δ_H 9.83–9.75 (1H, m, H-9′, H-9a′) were two formyl hydrogen signals. The ¹³C NMR data (Table 2) of 3 showed 10 carbon signals, including 2 methoxy hydrogen signals, 2 methene signals, 2 methenyl signals, 2 formyl signals, and 2 other sp3 hybrid carbon signals. The above HRMS and NMR data (Table 2; Supplementary Figure 12) indicated that 3 was polyunsaturated fatty acid structure (Figure 6C).

Compound 4 was isolated as a yellow powder and its molecular formula was deduced as C₄H₁₂N₂O on the basis of its UPLC-Q-Orbitrap HRMS data (m/z 104.0949), sharing several homologous fragment ions with AFB₁ such as 77, 69, 43 etc. (Figure 5E). The ¹H NMR data (Table 2; Supplementary Figure 15) of 4 showed the hydrogen signals of the aminomethyl group [δ_H 2.88–2.68 (2H, m, H-1‴), 1.43 (2H, t, J = 6.5 Hz, H-2‴)]. δ_H 1.82–1.51(2H, m, H-3‴, H-4‴) and δ_H 3.53 (2H, d, J = 11.7, H-5‴) were the hydrogen signals of methylen. δ_H 5.45 (2H, s, H-5a‴) was the aminogruppe hydrogen signal. The ¹³C NMR data (Table 2) of 4 showed 4 carbon signals, including 3 carbon signals on methylene signals, 1 aminomethyl group signal. The compound 4 was the 1-aminooxy-4-aminobutane (Figure 6D).

The structure of degradation products by the strain were determined using NMR as shown in Figure 6. The chemical structures of the products were primarily consisted of dimethoxyphenyl and enoic acid (compound 1), dimethylamino and ethyl carbamate (compound 2), polyunsaturated fatty acid (compound 3), and aminomethyl (compound 4). Among these four product structures, there were no structures that were associated with the high toxicity of AFB₁, including the furan ring double bond, coumarin lactone ring, and cyclopentenone ring.

Based on the structure of the degradation products, it could be inferred that B. amyloliquefaciens YUAD7 mainly degraded AFB₁ through secondary degradation (Figure 7). Primary degradation was achieved through hydrogenation and enzyme modification, directly cleaving the coumarin moiety (at positions 10, 11, and 15) and the cyclopentenone ring (at position 14) structures from the AFB₁ parent structure. Simultaneously, the modification disrupted the furan ring structure at positions 8 and 9, resulting in the formation of compound C₁₂H₁₄O₄. Moreover, an enzymatic modification added reactions that collided with the [N⁺H⁺] ion precursor bound, possibly forming the compound C₅H₁₂N₂O₂. Secondary degradation involved further decomposition of the primary degradation products C₁₂H₁₄O₄ and C₅H₁₂N₂O₂, primarily through the removal of the -CO moiety. In this process, the products underwent additional structural adjustments and cleavage, forming simpler compounds such as C₁₀H₁₄O₂ and C₄H₁₂N₂O.